Optimization Problem

hgwelec

Hello and congratulations for your work,



Consider the following problem representation :

val1,val2,val3...valN,fitness

where
val1...valN are Boolean (T/F) variables
avgIncome is a real number

Can RM be used for maximizing avgIncome subject to an 'optimal' subset of val1....valN variables?




Many Thanks for your help

Harry

TobiasMalbrecht

Hi Harry,

unfortunately I did not quite understand your problem. You mean you have examples with boolean/binominal attributes and a numerical label (capturing the income)? And now you want to find an attribute combination that optimally predicts the income? Then what learner do you intend to use and which performance measure? Or did I understand you completely wrong?

Regards,
Tobias

hgwelec

Hello Tobias,

You are right i guess i should be more specific. Basically i would like to know whether RM could be used in a Portfolio Optimization problem.. The variables that have T/F values suggest that a specific product (stock, bond etc) is present or not in a portfolio. Each line of the dataset represents a portfolio and the last attribute is the return on investment (ROI) of the specific portfolio, (plus some other attributes which are irrelevant to this discussion) .

Therefore we wish to maximize the ROI by selecting an optimal subset of stocks/bonds (variables var1,var.varN).


I have successfully used RM for other Financial Applications (and it works pretty well!) but i am not sure for this one....!


Hope it is more clear now and Thanks for your help



Harry